Management method, device, equipment and storage medium based on material product matching and automatic generation

Abstract

This invention belongs to the field of materials management technology and discloses a management method, device, equipment, and storage medium based on materials and products matching and automatic generation. The method includes: performing AI semantic parsing on descriptions of materials and products from heterogeneous sources to extract key parameters; constructing a dynamic product feature matrix based on the key parameters, and using the dynamic product feature matrix to perform multi-level matching with a supply database to obtain multiple candidate product matching schemes; determining a target product matching scheme from the multiple candidate product matching schemes based on a semantic similarity threshold; aligning the parameter specifications of the target product matching scheme through a parameter mapping engine; establishing a target mapping relationship between materials and products based on the processed target product matching scheme, and managing materials and products using the mapping relationship. Through the above method, the mapping relationship can be dynamically updated in real time, and the accuracy can be improved by utilizing the constructed mapping relationship.

Description

Technical Field

[0001] This invention relates to the field of materials management technology, and in particular to a management method, apparatus, equipment and storage medium based on materials and products matching and automatic generation. Background Technology

[0002] In the field of modern materials management, with the increasing complexity of global supply chains and the diversification of e-commerce platforms, the sources of material product descriptions are becoming increasingly heterogeneous. Different suppliers, different e-commerce platforms, and different internal enterprise systems often have significant differences in their descriptions of the same product. These differences are reflected in multiple aspects, such as parameter naming, unit usage, data format, and semantic expression.

[0003] Existing material and product matching and management systems mainly rely on manual rule configuration or simple keyword matching algorithms. Manual rule configuration requires customizing mapping rules for each heterogeneous source, which is not only costly to develop, but also difficult to adapt to the rapid access of new sources and dynamic rule updates. Furthermore, when faced with complex product descriptions and multi-level supply databases, it is prone to generating a large number of mismatches or missed matches.

[0004] The above content is only used to help understand the technical solution of the present invention and does not represent an admission that the above content is prior art. Summary of the Invention

[0005] The main objective of this invention is to provide a management method, apparatus, equipment, and storage medium based on material and product matching and automatic generation, aiming to solve the technical problems of low efficiency and poor accuracy in managing materials and products using manual rules.

[0006] To achieve the above objectives, the present invention provides a management method based on material and product matching and automatic generation, the management method comprising the following steps:

[0007] AI semantic analysis is used to extract key parameters from product descriptions of materials from heterogeneous sources.

[0008] A dynamic product feature matrix is ​​constructed based on the key parameters of the product, and the dynamic product feature matrix is ​​used to perform multi-level matching with the supply database to obtain multiple candidate product matching schemes.

[0009] The target product matching scheme is determined from multiple candidate product matching schemes based on semantic similarity threshold;

[0010] The parameter mapping engine is used to align the parameters of the target product matching scheme.

[0011] Based on the processed target product matching scheme, a target mapping relationship between materials and products is established, and the materials and products are managed using the mapping relationship.

[0012] In some embodiments, the AI ​​semantic parsing of the descriptions of materials and products from heterogeneous sources includes:

[0013] Construct an industry knowledge graph covering the bulk commodity sector;

[0014] Based on the industry knowledge graph, a domain-adaptive BERT model is trained, and the domain-adaptive BERT model is used to perform named entity recognition on unstructured product descriptions in order to extract different types of parameters.

[0015] The different types of parameters are parsed using the industry knowledge graph to generate a structured parameter tree with dependency tags;

[0016] The numerical nodes in the structured parameter tree are subjected to unit system fusion processing to obtain a standardized numerical vector;

[0017] Align the standardized numerical vector with the entities in the industry knowledge graph.

[0018] In some embodiments, the step of parsing the different types of parameters in conjunction with the industry knowledge graph to generate a structured parameter tree with dependency tags includes:

[0019] Based on the industry knowledge graph, the triple relationships of quantifiers, entities, and modifiers in product descriptions are annotated to generate weighted grammatical dependency templates.

[0020] Based on the grammatical dependency template, the nested structure of different types of parameters is parsed to generate a dependency tree with topological tags;

[0021] Perform context disambiguation on the ambiguous nodes in the dependency tree;

[0022] When conflicting parameters are identified among different types of parameters, source tracing and annotation are triggered, and an anomaly confidence score is generated.

[0023] A structured parametric tree with dependency labels is constructed by integrating the dependency tree topology, disambiguation results, and anomaly confidence scores.

[0024] In some embodiments, determining the target product matching scheme from multiple candidate product matching schemes based on a semantic similarity threshold includes:

[0025] Determine the semantic similarity between each candidate product matching scheme and the target material description;

[0026] The similarity threshold is dynamically adjusted based on the statistical distribution of historical matching data.

[0027] From multiple candidate product matching schemes, select candidate product matching schemes that exceed the similarity threshold;

[0028] From the candidate product matching schemes that exceed the similarity threshold, the candidate product scheme with the largest parameter dimension is selected as the target product matching scheme.

[0029] In some embodiments, the parameter specification alignment processing of the target product matching scheme through the parameter mapping engine includes:

[0030] The parameters from different sources in the target product matching scheme are converted into a unified parameter specification, and the parameters with different units in the target product matching scheme are converted into units so that parameters of the same type have the same unit.

[0031] Fill in the missing parameters in the processed target product matching scheme;

[0032] For parameters from different conflicting sources, the final parameter is determined according to a preset priority rule.

[0033] In some embodiments, the missing parameters in the filled target product matching scheme include:

[0034] Determine the first and second parameters adjacent to the missing parameter;

[0035] Determine the parameter number corresponding to the first parameter and the second parameter, as well as the parameter number corresponding to the missing parameter;

[0036] The parameter value corresponding to the missing parameter is calculated based on the parameter number corresponding to the first parameter and the second parameter, the parameter value corresponding to the first parameter and the second parameter, and the missing parameter is filled in based on the parameter value.

[0037] In some embodiments, establishing the target mapping relationship between materials and products based on the processed target product matching scheme includes:

[0038] Based on the parameter alignment results of the target matching scheme, triples containing material codes, product feature vectors and mapping weights are extracted, and reference mapping rules are generated by combining the process constraint relationships in the knowledge graph.

[0039] Analyze the weight conflicts between the reference mapping rules and the old mapping rules in the historical rule base;

[0040] If the weight conflict rate does not exceed the threshold, the material node and product node will be connected through multi-dimensional edges to construct the target mapping relationship. The edge attributes include matching degree score, parameter deviation threshold and effective time window.

[0041] Furthermore, to achieve the above objectives, the present invention also proposes a management device based on material product matching and automatic generation, the management device comprising:

[0042] The extraction module is used to perform AI semantic parsing on the descriptions of materials and products from heterogeneous sources to extract key parameters of the products;

[0043] The matching module is used to construct a dynamic product feature matrix based on the key parameters of the product, and to use the dynamic product feature matrix to perform multi-level matching with the supply database to obtain multiple candidate product matching schemes.

[0044] The filtering module is used to determine the target product matching scheme from multiple candidate product matching schemes based on a semantic similarity threshold.

[0045] The processing module is used to perform parameter specification alignment processing on the target product matching scheme through the parameter mapping engine;

[0046] The module is used to establish a target mapping relationship between materials and products based on the processed target product matching scheme, and to manage the materials and products using the mapping relationship.

[0047] Furthermore, to achieve the above objectives, the present invention also proposes a management device based on material product matching and automatic generation. The management device based on material product matching and automatic generation includes: a memory, a processor, and a management program based on material product matching and automatic generation stored in the memory and executable on the processor. The management program based on material product matching and automatic generation is configured to implement the steps of the management method based on material product matching and automatic generation as described above.

[0048] Furthermore, to achieve the above objectives, the present invention also proposes a computer program product, wherein the computer program product stores a management program based on material product matching and automatic generation, and the management program based on material product matching and automatic generation, when executed by a processor, implements the steps of the management method based on material product matching and automatic generation as described above.

[0049] This invention employs AI semantic parsing to extract key parameters from heterogeneous material product descriptions. A dynamic product feature matrix is ​​constructed based on these key parameters, and this matrix is ​​then used for multi-level matching with a supply database to obtain multiple candidate product matching schemes. A target product matching scheme is determined from these candidate schemes based on a semantic similarity threshold. The target product matching scheme undergoes parameter specification alignment processing using a parameter mapping engine. A target mapping relationship between materials and products is established based on the processed target product matching scheme, and this mapping relationship is used to manage the materials and products. This approach enables dynamic and real-time updates to the mapping relationship, while also improving accuracy. Attached Figure Description

[0050] Figure 1 This is a flowchart illustrating the first embodiment of the management method based on material and product matching and automatic generation of the present invention;

[0051] Figure 2 This is a structural block diagram of the first embodiment of the management device based on material product matching and automatic generation of the present invention.

[0052] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0053] It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the invention.

[0054] This invention provides a management method based on material and product matching and automatic generation, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of a management method based on material and product matching and automatic generation according to the present invention.

[0055] In this embodiment, the management method based on material product matching and automatic generation includes the following steps:

[0056] Step S10: Perform AI semantic parsing on the product descriptions of materials from heterogeneous sources to extract key product parameters.

[0057] In this embodiment, the execution entity is a management device based on material product matching and automatic generation. This management device has functions such as data processing, data communication, and program execution. The management device based on material product matching and automatic generation can be a computer terminal device or other network device, or other devices with similar functions. This embodiment does not limit the scope of the application.

[0058] It should be noted that existing material and product matching and management systems mainly rely on manual rule configuration or simple keyword matching algorithms. Manual rule configuration requires customizing mapping rules for each heterogeneous source, which is not only costly to develop but also difficult to adapt to the rapid integration of new sources and dynamic rule updates. Furthermore, when faced with complex product descriptions and multi-level supply databases, it is prone to generating numerous mismatches or missed matches.

[0059] In the specific implementation, for materials and products from heterogeneous sources, that is, materials and products from different sources, this embodiment adopts AI semantic parsing to extract the key parameters of the products. For example, if a company's product is 12mm, 12m, HRB400E rebar, after extracting the key parameters, we can obtain the product name: rebar, material: HRB400E, and specification: 12*12. This example is only for illustration. In actual applications, the parameter dimensions of the extracted key parameters can be set according to actual needs. This embodiment does not impose any restrictions on this.

[0060] Furthermore, in this embodiment, when performing IA semantic parsing, it is necessary to construct an industry knowledge graph covering the commodity domain; train a domain-adaptive BERT model based on the industry knowledge graph, and use the domain-adaptive BERT model to perform named entity recognition on unstructured product descriptions to extract different types of parameters; parse the different types of parameters in conjunction with the industry knowledge graph to generate a structured parameter tree with dependency tags; perform unit system fusion processing on the numerical nodes in the structured parameter tree to obtain a standardized numerical vector; and align the standardized numerical vector with the entities in the industry knowledge graph.

[0061] It should be noted that the industry knowledge graph includes product categories, parameter types, unit conversion relationships, etc. This industry knowledge graph is used to load national standard parameter systems and establish a three-dimensional entity relationship network of materials, processes, and specifications. Training the domain-adaptive BERT model using the industry knowledge graph allows the knowledge graph to provide entity relationship constraints for the BERT model, ensuring that named entity recognition conforms to industry standards. The parameters extracted by the domain-adaptive BERT model include, but are not limited to, material categories, geometric specifications, and process parameters. The unit system fusion processing in this embodiment targets metric conversions for imperial units (e.g., feet) and non-standard units (e.g., pieces / bundles), resulting in standardized numerical vectors. Entity alignment essentially compares the standardized numerical vectors with entities in the knowledge graph. A discriminator identifies regional differences in description; for example, wire rod and wire are regionally different descriptions, but they actually refer to the same material. Entity alignment can unify the references to materials or products with regionally different descriptions.

[0062] In one embodiment, the step of parsing the different types of parameters using the industry knowledge graph to generate a structured parameter tree with dependency tags further includes: generating a weighted syntactic dependency template based on the triple relationships of quantifiers, entities, and modifiers in the product description labeled by the industry knowledge graph; parsing the nested structure of different types of parameters based on the syntactic dependency template to generate a dependency tree with topological tags; performing context disambiguation on ambiguous nodes in the dependency tree; triggering source tracing and generating anomaly confidence scores when conflicting parameters are identified among different types of parameters; and integrating the dependency tree topology, disambiguation results, and anomaly confidence scores to establish a structured parameter tree with dependency tags.

[0063] It should be noted that nested relationships, such as identifying the nested relationship between parameters like diameter Φ and length L (e.g., Φ12*12m), can be addressed through context disambiguation. This can be achieved by using a multi-head attention mechanism to analyze the semantic meaning of connectives (and / inclusive), combined with a knowledge graph-based process attribute database, to resolve the actual referents of polysemous parameters (e.g., the parallel relationship in galvanizing and coating). In practice, conflicting parameters may exist, such as HRB400E stainless steel. In this case, anomaly confidence scoring is performed. Finally, after fusing the dependency tree topology, disambiguation results, and anomaly confidence scores, a four-tuple sequence is output, consisting of an entity, attribute, value, and confidence level.

[0064] Step S20: Construct a dynamic product feature matrix based on the key parameters of the product, and use the dynamic product feature matrix to perform multi-level matching with the supply database to obtain multiple candidate product matching schemes.

[0065] In the specific implementation, after extracting the above key parameters, this embodiment uses the key parameters to perform multi-level matching with the supply database. The multi-level matching in this embodiment refers to matching different parameter dimensions. For example, first match the material name, then match the material material, and finally match the material specifications. Since there are many similar products in actual situations, especially products of similar types, multiple candidate product matching schemes can be obtained after matching the above key parameters. Each matching scheme corresponds to a different product.

[0066] Step S30: Determine the target product matching scheme from multiple candidate product matching schemes based on the semantic similarity threshold.

[0067] In this embodiment, after obtaining the above multiple candidate product matching schemes, it is necessary to determine the semantic similarity between each candidate product matching scheme and the description of the target material; dynamically adjust the similarity threshold according to the statistical distribution of historical matching data; filter out candidate product matching schemes that exceed the similarity threshold from the multiple candidate product matching schemes; and determine the candidate product scheme with the largest parameter dimension from the candidate product matching schemes that exceed the similarity threshold as the target product matching scheme.

[0068] It should be noted that the similarity threshold can be adjusted, for example, θ = μ + k·σ, where μ represents the mean semantic similarity of historical matching data, σ represents the standard deviation, and k is a preset confidence coefficient. The final target product matching scheme corresponds to the largest parameter dimension. By designing the maximum parameter dimension, the effectiveness of the subsequently established material-product mapping relationship can be improved, thereby increasing accuracy.

[0069] Step S40: Perform parameter specification alignment processing on the target product matching scheme through the parameter mapping engine.

[0070] In practical implementation, after determining the target product matching scheme, to make the scheme more standardized, parameters from different sources are converted into a unified parameter specification. Simultaneously, parameters with different units are converted to ensure that parameters of the same type have the same unit. This unit unification is achieved, for example, meters are uniformly converted to millimeters. Furthermore, this embodiment can also complete missing parameters. The specific completion process involves: determining the first and second parameters adjacent to the missing parameter; determining the parameter indices corresponding to the first and second parameters, and the parameter indice corresponding to the missing parameter; calculating the parameter value corresponding to the missing parameter based on the parameter indices, parameter values, and parameter indices, and then filling in the missing parameter based on the parameter value.

[0071] It should be noted that the formula for calculating the parameter value corresponding to the missing parameter mentioned above is as follows:

[0072]

[0073] Where, x i and x i+1 These are the parameter values ​​corresponding to the first and second parameters, respectively. i and t i+1 These are the parameter numbers corresponding to the first and second parameters, respectively. m x is the parameter index corresponding to the missing parameter. m That is, the parameter value that is missing.

[0074] Furthermore, in practice, there may be parameter conflicts. Parameters of the same type may have different sources, resulting in different values. For example, there may be two specification ranges for the same material. In this case, this embodiment can use a preset priority rule to determine the final parameter. For example, the parameter within the parameter value range can be given the highest priority, or the parameter with the smaller value can be given the highest priority.

[0075] Step S50: Establish a target mapping relationship between materials and products based on the processed target product matching scheme, and use the mapping relationship to manage the materials and products.

[0076] It should be noted that the mapping rules, i.e., the mapping relationship between materials and products, are updated in real time. Therefore, determining the new mapping relationship requires analyzing the weight conflicts with the old mapping rules in the historical rule base. Furthermore, it should be understood that the changes in material and product parameters are real-time and do not exhibit significant deviations. Therefore, the conflict rate between the corresponding new mapping relationship and the old mapping rules in the historical rule base should not exceed a threshold. In this embodiment, the threshold can be set to 10%, but it can be set to other values ​​according to actual needs; this embodiment does not impose any restrictions on this. The reference mapping rule refers to the mapping relationship between the triplet of material code, product feature vector, and mapping weight.

[0077] In a practical implementation, weight conflict can be calculated by the difference between the mapping weight of the reference mapping rule and the mapping weight of the old mapping rule in the historical rule library. Taking the above threshold as an example, if the difference does not exceed 10% of the mapping weight of the old mapping rule in the historical rule library, the conflict rate is considered to meet the requirements.

[0078] In this embodiment, AI semantic parsing is performed on the descriptions of materials and products from heterogeneous sources to extract key parameters of the products. A dynamic product feature matrix is ​​constructed based on the key parameters of the products, and the dynamic product feature matrix is ​​used to perform multi-level matching with the supply database to obtain multiple candidate product matching schemes. The target product matching scheme is determined from the multiple candidate product matching schemes according to the semantic similarity threshold. The parameter specification alignment processing of the target product matching scheme is performed through a parameter mapping engine. Based on the processed target product matching scheme, a target mapping relationship between materials and products is established, and the materials and products are managed using the mapping relationship. Through the above method, the mapping relationship can be dynamically updated in real time, and the accuracy can be improved by using the constructed mapping relationship.

[0079] Furthermore, this invention also proposes a computer program product, which stores a management program based on material product matching and automatic generation. When the management program based on material product matching and automatic generation is executed by a processor, it implements the steps of the management method based on material product matching and automatic generation described above.

[0080] Reference Figure 2 , Figure 2 This is a structural block diagram of the first embodiment of the management device based on material product matching and automatic generation of the present invention.

[0081] like Figure 2 As shown, the management device based on material product matching and automatic generation proposed in this embodiment of the invention includes:

[0082] Extraction module 10 is used to perform AI semantic parsing on the descriptions of materials and products from heterogeneous sources and extract key parameters of the products;

[0083] The matching module 20 is used to construct a dynamic product feature matrix based on the key parameters of the product, and to use the dynamic product feature matrix to perform multi-level matching with the supply database to obtain multiple candidate product matching schemes.

[0084] The filtering module 30 is used to determine the target product matching scheme from multiple candidate product matching schemes based on a semantic similarity threshold.

[0085] Processing module 40 is used to perform parameter specification alignment processing on the target product matching scheme through the parameter mapping engine;

[0086] The construction module 50 is used to establish a target mapping relationship between materials and products based on the processed target product matching scheme, and to manage the materials and products using the mapping relationship.

[0087] In this embodiment, AI semantic parsing is performed on the descriptions of materials and products from heterogeneous sources to extract key parameters of the products. A dynamic product feature matrix is ​​constructed based on the key parameters of the products, and the dynamic product feature matrix is ​​used to perform multi-level matching with the supply database to obtain multiple candidate product matching schemes. The target product matching scheme is determined from the multiple candidate product matching schemes according to the semantic similarity threshold. The parameter specification alignment processing of the target product matching scheme is performed through a parameter mapping engine. Based on the processed target product matching scheme, a target mapping relationship between materials and products is established, and the materials and products are managed using the mapping relationship. Through the above method, the mapping relationship can be dynamically updated in real time, and the accuracy can be improved by using the constructed mapping relationship.

[0088] In some embodiments, the extraction module 10 is used to construct an industry knowledge graph covering the commodity sector;

[0089] Based on the industry knowledge graph, a domain-adaptive BERT model is trained, and the domain-adaptive BERT model is used to perform named entity recognition on unstructured product descriptions in order to extract different types of parameters.

[0090] The different types of parameters are parsed using the industry knowledge graph to generate a structured parameter tree with dependency tags;

[0091] The numerical nodes in the structured parameter tree are subjected to unit system fusion processing to obtain a standardized numerical vector;

[0092] Align the standardized numerical vector with the entities in the industry knowledge graph.

[0093] In some embodiments, the extraction module 10 is used to generate a weighted grammatical dependency template based on the triple relationships of quantifiers, entities, and modifiers in the product description labeled by an industry knowledge graph.

[0094] Based on the grammatical dependency template, the nested structure of different types of parameters is parsed to generate a dependency tree with topological tags;

[0095] Perform context disambiguation on the ambiguous nodes in the dependency tree;

[0096] When conflicting parameters are identified among different types of parameters, source tracing and annotation are triggered, and an anomaly confidence score is generated.

[0097] A structured parametric tree with dependency labels is constructed by integrating the dependency tree topology, disambiguation results, and anomaly confidence scores.

[0098] In some embodiments, the filtering module 30 is used to determine the semantic similarity between each candidate product matching scheme and the target material description;

[0099] The similarity threshold is dynamically adjusted based on the statistical distribution of historical matching data.

[0100] From multiple candidate product matching schemes, select candidate product matching schemes that exceed the similarity threshold;

[0101] From the candidate product matching schemes that exceed the similarity threshold, the candidate product scheme with the largest parameter dimension is selected as the target product matching scheme.

[0102] In some embodiments, the processing module 40 is used to convert parameters from different sources in the target product matching scheme into a unified parameter specification, and to convert parameters of different units in the target product matching scheme into units so that parameters of the same type have the same unit.

[0103] Fill in the missing parameters in the processed target product matching scheme;

[0104] For parameters from different conflicting sources, the final parameter is determined according to a preset priority rule.

[0105] In some embodiments, the processing module 40 is used to determine a first parameter and a second parameter adjacent to the missing parameter;

[0106] Determine the parameter number corresponding to the first parameter and the second parameter, as well as the parameter number corresponding to the missing parameter;

[0107] The parameter value corresponding to the missing parameter is calculated based on the parameter number corresponding to the first parameter and the second parameter, the parameter value corresponding to the first parameter and the second parameter, and the missing parameter is filled in based on the parameter value.

[0108] In some embodiments, the construction module 50 is used to extract a triple containing material code, product feature vector and mapping weight based on the parameter alignment result of the target matching scheme, and generate a reference mapping rule in combination with the process constraint relationship in the knowledge graph.

[0109] Analyze the weight conflicts between the reference mapping rules and the old mapping rules in the historical rule base;

[0110] If the weight conflict rate does not exceed the threshold, the material node and product node will be connected through multi-dimensional edges to construct the target mapping relationship. The edge attributes include matching degree score, parameter deviation threshold and effective time window.

[0111] This application embodiment also provides a management device based on material product matching and automatic generation, including a processor, a communication interface, a memory, and a communication bus. The processor, communication interface, and memory communicate with each other through the communication bus. The memory is used to store management programs based on material product matching and automatic generation. When the processor executes the program stored in the memory, it implements the above-mentioned management method based on material product matching and automatic generation.

[0112] The communication bus mentioned in the aforementioned management equipment based on material product matching and automatic generation can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. This communication bus can be divided into address bus, data bus, control bus, etc.

[0113] The communication interface is used for communication between the aforementioned management equipment based on material product matching and automatic generation and other equipment.

[0114] The memory may include random access memory (RAM) or non-volatile memory (NVM), such as at least one disk storage device. Optionally, the memory may also be at least one storage device located remotely from the aforementioned processor.

[0115] The processors mentioned above can be general-purpose processors, including central processing units (CPUs), network processors (NPs), etc.; they can also be digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0116] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid state disk (SSD)).

[0117] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply the existence of any such actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or device comprising the element.

[0118] The various embodiments in this specification are described in a related manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0119] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

[0120] It should be understood that the above are merely illustrative examples and do not constitute any limitation on the technical solutions of the present invention. In specific applications, those skilled in the art can make settings as needed, and the present invention does not impose any restrictions on this.

[0121] It should be noted that the workflow described above is merely illustrative and does not limit the scope of protection of this invention. In practical applications, those skilled in the art can select some or all of the workflow to achieve the purpose of this embodiment according to actual needs, and no restrictions are imposed here.

[0122] In addition, for technical details not described in detail in this embodiment, please refer to the management method based on material product matching and automatic generation provided in any embodiment of the present invention, which will not be repeated here.

[0123] Furthermore, it should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or system. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or system that includes that element.

[0124] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0125] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as read-only memory (ROM) / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0126] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

[0127] It is understood that the system provided in the embodiments of the present invention corresponds to the method provided in the embodiments of the present invention, and the explanation, examples and beneficial effects of the relevant content can be referred to the corresponding parts of the above methods.

Claims

1. A management method based on material and product matching and automatic generation, characterized in that, The management method based on material product matching and automatic generation includes: AI semantic analysis is used to extract key parameters from product descriptions of materials from heterogeneous sources. The AI ​​semantic parsing of descriptions of materials and products from heterogeneous sources includes: Construct an industry knowledge graph covering the bulk commodity sector; Based on the industry knowledge graph, a domain-adaptive BERT model is trained, and the domain-adaptive BERT model is used to perform named entity recognition on unstructured product descriptions in order to extract different types of parameters. Based on the industry knowledge graph, the triple relationships of quantifiers, entities, and modifiers in product descriptions are annotated to generate weighted grammatical dependency templates. Based on the grammatical dependency template, the nested structure of different types of parameters is parsed to generate a dependency tree with topological tags; Perform context disambiguation on the ambiguous nodes in the dependency tree; When conflicting parameters are identified among different types of parameters, source tracing and annotation are triggered, and an anomaly confidence score is generated. A structured parametric tree with dependency labels is constructed by integrating the dependency tree topology, disambiguation results, and anomaly confidence scores. The numerical nodes in the structured parameter tree are subjected to unit system fusion processing to obtain a standardized numerical vector; Align the standardized numerical vector with entities in the industry knowledge graph; A dynamic product feature matrix is ​​constructed based on the key parameters of the product, and the dynamic product feature matrix is ​​used to perform multi-level matching with the supply database to obtain multiple candidate product matching schemes. The target product matching scheme is determined from multiple candidate product matching schemes based on semantic similarity threshold; The parameter mapping engine is used to align the parameters of the target product matching scheme. Based on the processed target product matching scheme, a target mapping relationship between materials and products is established, and the mapping relationship is used to manage the materials and products. The establishment of a target mapping relationship between materials and products based on the processed target product matching scheme includes: Based on the parameter alignment results of the target matching scheme, triples containing material codes, product feature vectors and mapping weights are extracted, and reference mapping rules are generated by combining the process constraint relationships in the knowledge graph. Analyze the weight conflicts between the reference mapping rules and the old mapping rules in the historical rule base; If the weight conflict rate does not exceed the threshold, the material node and product node will be connected through multi-dimensional edges to construct the target mapping relationship. The edge attributes include matching score, parameter deviation threshold and effective time window.

2. The management method based on material product matching and automatic generation as described in claim 1, characterized in that, The step of determining the target product matching scheme from multiple candidate product matching schemes based on a semantic similarity threshold includes: Determine the semantic similarity between each candidate product matching scheme and the target material description; The similarity threshold is dynamically adjusted based on the statistical distribution of historical matching data. From multiple candidate product matching schemes, select candidate product matching schemes that exceed the similarity threshold; From the candidate product matching schemes that exceed the similarity threshold, the candidate product scheme with the largest parameter dimension is selected as the target product matching scheme.

3. The management method based on material product matching and automatic generation as described in claim 1, characterized in that, The parameter specification alignment process for the target product matching scheme using a parameter mapping engine includes: The parameters from different sources in the target product matching scheme are converted into a unified parameter specification, and the parameters with different units in the target product matching scheme are converted into units so that parameters of the same type have the same unit. Fill in the missing parameters in the processed target product matching scheme; For parameters from different conflicting sources, the final parameter is determined according to a preset priority rule.

4. The management method based on material product matching and automatic generation as described in claim 3, characterized in that, The missing parameters in the target product matching scheme after filling in the missing parameters include: Determine the first and second parameters adjacent to the missing parameter; Determine the parameter number corresponding to the first parameter and the second parameter, as well as the parameter number corresponding to the missing parameter; The parameter value corresponding to the missing parameter is calculated based on the parameter number corresponding to the first parameter and the second parameter, the parameter value corresponding to the first parameter and the second parameter, and the missing parameter is filled in based on the parameter value.

5. A management device based on material product matching and automatic generation, characterized in that, The management device based on material product matching and automatic generation includes: The extraction module is used to perform AI semantic parsing on the descriptions of materials and products from heterogeneous sources to extract key parameters of the products; The extraction module is further used to construct an industry knowledge graph covering the bulk commodity sector; Based on the industry knowledge graph, a domain-adaptive BERT model is trained, and the domain-adaptive BERT model is used to perform named entity recognition on unstructured product descriptions in order to extract different types of parameters. Based on the industry knowledge graph, the triple relationships of quantifiers, entities, and modifiers in product descriptions are annotated to generate weighted grammatical dependency templates. Based on the grammatical dependency template, the nested structure of different types of parameters is parsed to generate a dependency tree with topological tags; Perform context disambiguation on the ambiguous nodes in the dependency tree; When conflicting parameters are identified among different types of parameters, source tracing and annotation are triggered, and an anomaly confidence score is generated. A structured parametric tree with dependency labels is constructed by integrating the dependency tree topology, disambiguation results, and anomaly confidence scores. The numerical nodes in the structured parameter tree are subjected to unit system fusion processing to obtain a standardized numerical vector; Align the standardized numerical vector with entities in the industry knowledge graph; The matching module is used to construct a dynamic product feature matrix based on the key parameters of the product, and to use the dynamic product feature matrix to perform multi-level matching with the supply database to obtain multiple candidate product matching schemes. The filtering module is used to determine the target product matching scheme from multiple candidate product matching schemes based on a semantic similarity threshold. The processing module is used to perform parameter specification alignment processing on the target product matching scheme through the parameter mapping engine; The module is used to establish a target mapping relationship between materials and products based on the processed target product matching scheme, and to manage the materials and products using the mapping relationship; The construction module is further used to extract triples containing material codes, product feature vectors and mapping weights based on the parameter alignment results of the target matching scheme, and generate reference mapping rules by combining the process constraint relationships in the knowledge graph. Analyze the weight conflicts between the reference mapping rules and the old mapping rules in the historical rule base; If the weight conflict rate does not exceed the threshold, the material node and product node will be connected through multi-dimensional edges to construct the target mapping relationship. The edge attributes include matching score, parameter deviation threshold and effective time window.

6. A management device based on material and product matching and automatic generation, characterized in that, The management device based on material product matching and automatic generation includes: a memory, a processor, and a management program based on material product matching and automatic generation stored on the memory and executable on the processor, wherein the management program based on material product matching and automatic generation is configured to implement the steps of the management method based on material product matching and automatic generation as described in any one of claims 1 to 4.

7. A computer program product, characterized in that, The computer program product stores a management program based on material product matching and automatic generation. When the management program based on material product matching and automatic generation is executed by the processor, it implements the steps of the management method based on material product matching and automatic generation as described in any one of claims 1 to 4.

Images